Detecting Online Fake Reviews Using Supervised and Semi Supervised Learning

Online reviews play a crucial role in influencing consumer decisions, which makes them a target for manipulation through fake or misleading feedback. Detecting such deceptive reviews is challenging because fraudulent content is often written to closely resemble genuine opinions. This project presents a hybrid approach for *detecting online fake reviews using supervised and semisupervised machine learning techniques*. The system abelled both linguistic features and reviewer behavior to classify reviews as genuine or deceptive. Supervised learning models such as Support Vector Machines, Logistic Regression, and Random Forest are trained on abelled datasets to establish a strong baseline. To address the scarcity of high-quality abelled data, semisupervised methods—including Self-Training and Label Propagation—are integrated to utilize large amounts of unlabeled reviews. This combination enhances model robustness and improves detection accuracy in real-world scenarios. Experimental results demonstrate that the semi-supervised models significantly improve performance, especially when abelled data is limited. The proposed hybrid approach offers an effective and scalable solution for identifying deceptive content, helping e-commerce platforms protect customers and maintain trust.